IMAGE REFORMATION TECHNIQUES IN CT-SABBU.pptx

Image reformation techniques in CT

1. IMAGE REFORMATION TECHNIQUES IN CT PRESENTED BY:-SABBU KHATOON Roll no: 156 BSc MIT 2 nd year MMC,IOM

2. Presentation Layout Introduction to Image Reformation Terminologies Reconstruction VS Reformation Requirements for Image Reformation Two dimensional image formats Three dimensional Imaging Three dimensional image formats Significance and drawbacks of Image Reformation Factors that degrade reformatted image Summary References

3. INTRODUCTION Image reformation is also known as image rendering ; is a image post processing technique in CT Image data are assembled to produce images in different planes, or to produce 3D images It is essential to have the original data of excellent quality, to perform image reformatting Better display anatomic relationships Modern scanner allow many opportunities to manipulate both raw data and image data

4. Terminologies Raw data:- an initial measurement or projections obtained by the CT scanner after scanning the patient. Image data:- the processed and reconstructed form of raw data after applying mathematical transformations and algorithms. DFOV:- (zoom/ target) determines how much of the collected raw data is used to create an image SFOV:- determines the area, within the gantry, from which the raw data are acquired ; determines the number of detector cells collecting data

5. Image center:- it is the geometric midpoint of the reconstructed CT image, representing the center of the field of view (FOV). Prospective reconstruction:- reconstruction that is automatically produced during scanning Retrospective reconstruction:- using the same raw data later to generate a new image Anisotropic voxel:- A voxel with unequal dimensions in the three axes, often with thinner slices in one dimension. Isotropic voxel:- A voxel with equal dimensions in all three axes (x, y, z). Facilitates accurate 3D reconstructions and multiplanar reformation (MPR).

6. ASPECTS IMAGE RECONSTRUCTION IMAGE REFORMATION Definition Process of creating 2D image from raw data Process of generating new views or format from reconstructed image Primary purpose Generate initial cross- sectional image Provides different perspectives and formats for analysis Techniques Used Filtered back projection, Iterative reconstruction MPR, MIP, MinIP , VRT Data Source From Raw data From Image data Output 2D axial slices 2D/3D reformatted images Impact on Image Quality Direct influence on quality and noise levels Dependent on initial reconstruction quality Flexibility Limited by reconstruction parameters High, multiple formats and views Applications In- plane imaging, internal structure identification Diagnostic imaging, advanced diagnostic visualization Reconstruction vs Reformation

7. Requirements For Image Reformation For CT image reformation, all the source images must have an identical : DFOV, image center, gantry tilt they must be contiguous

8. CT Image Reformatting Techniques Two- dimensional formats Multiplanar reformation (MPR) Curved planar reformation (CPR) Three- dimensional formats Surface shaded displays (SSD) Volume rendering (VR) Average intensity projection (AIP) Maximum intensity projection (MIP) Minimum intensity projection (MinIP) Endoluminal imaging Cinematic rendering

9. Multiplanar Reformation (MPR) Reformation that is done to show anatomy in various planes is referred to as multiplanar reformation (MPR). 2D in nature can create coronal, sagittal, and paraxial images from a stack of contiguous transverse axial scans If voxels are isotropic, the reformatted image is virtually identical in quality to the original axial images If voxels are anisotropic then the image quality can be improved by using overlapping images

10. Diagrammatic representation of MPR

11. MPR in Coronal Plane MPR in Sagittal Plane MPR in Oblique Plane

12. Curved Planar Reformation Also called curved multi- planar reformation or cMPR Allows Vessel- Tracking CPR allows the images to be created along the centerline of tubular organs (eg vessels, CBD, ureters etc.) in a single image. FIG:- CPR for vessel visualization

13. CPR and dental CBCT could identify more dental pathologies than conventional CT images. (A) The parasagittal CT view identified a periapical lesion of a left maxillary molar however, both (B) CPR and (C) CBCT revealed a co-existing fistula

14. Unfolded Cylindrical Projection (UCP) Reformat Anatomy of ribcage makes CT evaluation a time consuming process because no orthogonal plane is optimal for visualization of all the ribs of an individual.  This technique automatically creates a 2D cylindrical projection of the whole rib cage based on a complex 3D manifold system following the long axis of the ribs 

15. Three- Dimensional Imaging Is a method by which a set of data is collected from a 3D object , processed by a computer, and displayed on a 2D computer screen to give the illusion of depth. Depth perception causes the image to appear in three dimensions Provides both qualitative and quantitative information from images to aid diagnosis Qualitative information is used to compare how observers perform on a specific task to demonstrate the diagnostic value of 3D imaging Quantitative information is used to assess three elements of the technique: precision (reliability), accuracy (true detection), and efficiency (feasibility) of the 3D imaging procedure

16. Fundamental 3D Concepts Four co-ordinate systems - Scanner coordinate system; abc - Display coordinate system; rst - Object coordinate system; uvw - Scene coordinate system; xyz Most familiar is scene coordinate system. X=width of object Y= height of object Z= depth dimension that adds realism to image The xyz coordinate define a space in which multidimensional data (set of slices) are represented known as 3D space or scene space. Graphic representation of 4 coordinate system used in 3D imaging

17. Steps In Creating 3D Images

18. Processing For 3D Imaging Segmentation Thresholding Object delineation Rendering

19. Segmentation Also known as, region of interest editing, is a processing techniques used to identify the structure of interest in a given scene and determines the voxels to be either displayed or discarded Manual, semi automatic or fully automatic segmentation It determines which voxels are a part of the object and which are not, and should be discarded.

20. Segmentation Manual segmentation technique. Image (A) shows the CT head holder. In image (B), the technologist traced the right side of the head holder for removal.

21. Thresholding a process that use one or more threshold or CT number ranges to define the different volumes that will undergo segmentation. CT number is used to determine this.

22. Thresholding (In Segmentation) Applying thresholding on the CT image. (a) Original CT images. (b) Extracting the bone by using thresholding segmentation

23. Object Delineation Is portraying an object by involving boundaries and volume extraction and detection methods Boundary extraction:- focus on identifying voxels that define outer/inner borders/surfaces of an object. Results in Object contour. Volume extraction: - Aims to identify all the voxels within the 3D space and on the surface of the object

24. Rendering Is a stage when an image in 3D space is transformed into a simulated 3D image to be displayed on 2D computer screen Three rendering techniques: Intensity projection rendering Surface rendering Volume rendering

25. Intensity Projection Rendering A 3D volume is represented as a stack of 2D image slices, with each slice containing pixel values that correspond to intensity or density of the underlying structures an extension of MPR techniques and consist of generating arbitrary thick slices (slabs) from thin slices. Method of image projection ; Ray tracing ; a mathematical ray from the observer's eye passes through the 2D screen and 3D volume, with the pixel intensity being the average of all intersected voxel intensities FIG:- Technique of Ray Tracing

26. Intensity Projection Rendering The term sliding thin slabs is known for this technique. The algorithms for sliding thin slabs (STS) include Average intensity projection (AIP), Maximum intensity projection (MIP), and Minimum intensity projection (MinIP)

27. Average Intensity Projection AIP images represents average attenuation values within the voxel With AIP images is achieved, an appearance similar to traditional axial low contrast resolution. This may be useful for the characterization of the internal structures of a solid body or walls of hollow structures such as blood vessels or intestine

28. Maximum Intensity Projection ( MIP ) MIP images show only the highest attenuation values within the voxels and especially useful to create angiographic and urography images Enables the detection of highly intense structure. MIP displays the higher CT number in a volume of interest when projected into a new plane. Two significant limitations of MIP are: The presence of other high-attenuating voxels may obscure evaluation of vasculature 3D relationships among the structures in the display are not visible

29. Raw data (stacks of 2D axial slices); each slice contains pixel intensity values corresponding to density of structure of body. For each ray, the ray tracing algorithm identifies the maximum intensities value. The result is a single 2D image where each pixel’s value is highest intensity encountered along the corresponding ray. Maximum Intensity Projection

30. Uses of MIP MIP is mainly used to show the vessels with contrast material in CT angiography to provide clear view of lesions

31. Thick and Thin-Slab MIP The thickness of the volume of interest can be modified to include or exclude various object from the projection Thick-slab MIPs incorporates a larger number of consecutive CT slices. Allows for a broader overview; used for larger structure . Useful in visualizing larger structures like major blood vessels, abdominal structures, etc. Thin slab MIP uses a small number of contiguous CT slices; provides high detail and used for small structures. Example:- examining fine detailed structures such as tiny lesions

32. Minimum Intensity Projection ( MinIP ) MinIP images shows only the lowest attenuation values. It enables detection of low- density structures. 2D image of a selected volume is generated where each pixel is represented by displaying the lowest attenuation value in each voxel. It is used to represent structures containing air such as tracheobronchial tree

33. Use of Minimum Intensity Projection It is mainly used to diagnose lung diseases. For example:- traction of bronchiectasis and emphysema

34. MinIP vs AIP vs MIP MinIP AIP MIP

35. Surface Rendering Aka shaded surface display (SSD) ; is a virtual 3D image. Displays the image according to it’s calculations of how the light rays would be reflected to the viewer’s eyes. Thresholding determines which densities will be displayed. SSD uses less than 10% of original data in 3D plane. Density information is lost Involves 2 steps: Surface formation and Rendering

36. Surface Rendering Surface formation involves the operation of contouring by using segmentation obtained by thresholding. The threshold setting determines whether skin or bone surfaces will be displayed Rendering follows surface formation and is intended to add photorealism and create the illusion of depth in an image, making it appear 3D on a 2D computer screen FIG:-Surface-rendered 3D image of the foot

37. Volume Rendering (VRT) Is a 3D semitransparent representation of the imaged structure. It undergoes following steps:- Selecting the viewing orientation and hypothetical screen. Casting rays through the 3D volume. Using a mapping function to determine the pixel values based on the intersections of the rays with the volume. Re-interpolating the 3D volume to align with the rays for computational efficiency FIG:-Geometric relationships of volume rendering.

38. CONTD.. Two stages to volume rendering Preprocessing volume Rendering Preprocessing the volume involves several image processing operations including segmentation that determine the tissue types contained in each voxel and to assign different brightness level or color Rendering involves image projection (ray tracing) to form the simulated 3D image Volume rendering offers the advantage of seeing through surfaces allowing us to examine both external and internal structures

39. VRT Images

40. 3D VRT of carotid arteries and its branches

41. Pros And Cons of VRT Images Pros: Enhanced 3D visualization of anatomic structure that aids to show spatial relationships between different structures and view them from various angles Valuable in surgical planning allowing surgeons to visualize the surgical site in 3D and determine the best approach Non invasive alternate for certain diagnostic and planning techniques Cons: Longer processing time Needs high computational power Artifacts may arise from various sources such as image acquisition, recon algorithms etc

42. Volume Rendering versus MIP Schematic comparison of volume-rendered and MIP images.  (a)  Volume-rendered image provides clear definition of individual vessels.  (b)  MIP image reconstructed from the same volume data shows all of the vessels, but their outlines merge; it is impossible to visualize the spatial relationships between the vessels or to delineate individual vessels on the MIP image. (a) (b)

43. MIP and VR images. In MIP images, hyperintense structures are superimposed and the three-dimensional perception is lost

44. Endoluminal Imaging A form of volume rendering (VR) Also called perspective volume rendering or virtual endoscopy Virtual bronchoscopy, virtual colonoscopy or CT colonography Designed to look inside the lumen of a structure.

45. Cinematic Rendering Also known as, global illumination rendering (Canon Medical Systems) or volume illumination (GE Healthcare) Is a novel 3D rendering algorithm that simulates the complete interaction of photons with the scanned object, providing photorealistic volume rendering The algorithm uses Monte Carlo path tracing method to generate hyper realistic images by light transport along hundreds or thousands of photons per pixel through the anatomy using a stochastic process Recently adopted in medical imaging

46. Advantages of Cinematic Rendering over VRT Physically correct simulation of light and shadow, allowing for better and immediate perception of the shapes, contours, location and relationship of various objects High quality denoising combined with edge preservation A minor limitation is that shadows can occlude areas

47. Contrast-enhanced 3-D CT images of a cervical spine injury with luxation of C3/4. Both CR (a) and VR (b) images show the anterior subluxation of the cervical vertebrae (arrows). The enhanced lightning of CR creates a high contrast between vessels and bone, making it easy to follow the course of the vertebral artery

48. 3D Rendering For best results in 3D rendering, Use thin slice Overlapping intervals Smoothing algorithms (to reduce image noise)

49. Significance of Image Reformation Provides multiple views (sagittal, coronal, and axial) for comprehensive analysis; Allows for better assessment of complex anatomical structures. 3D rendered image provide views of the imaging volume from different angles MIP and MinIP techniques can highlight high and low-density structures; improving detection of subtle abnormalities. Assists in radiotherapy planning by accurately delineating target volumes and critical structures

50. Drawbacks of Image reformatting Loss of image detail Reformation can introduce artifacts, particularly if the original scan data is not optimal. Reformatted images can increase the amount of data to be stored and managed Additional software and hardware capabilities required for advanced reformation techniques can be costly

51. Factors That Degrade Reformatted Image Segmentation error Image noise Artifact Motion Metal Stair step

52. Segmentation Errors Especially common when automatic segmentation and vessel tracking techniques are used Occurs when important vessels or other structures are inadvertently edited out of the data set Affected by 4 basic types of error viz added region, added background, inside holes and border holes which in turn may result in errors in boundary delineation and in determining the volume of segmented object Solution: Never depend solely on segmented images. Always review unsegmented volumes or MIPs and the thin section data

53. Segmentation Errors

54. Image Noise Excessive image noise in the source images will significantly limit the quality and utility of 3D rendered images. Common problem in very large patients. Excessive noise reduces image quality particularly its low contrast resolution Particularly problematic for vascular examinations cause- the contrast between the soft tissues and the contrast filled vessels is reduced Automatic segmentation techniques also tend to work poorly in this situation.

55. Image Noise Remedies: Appropriate scan parameters (high mAs, lower pitch setting, slower rotation speed) Larger DFOVs and thicker slices in case of obese patient

56. Artifacts Artifacts on the source data will also degrade the reformatted image Motion artifacts result from patient movement, breathing, pulsation or peristalsis They are particularly dangerous on multiplanar reconstructions. Subtle motion can easily mimic a fracture or a vascular lesion on reconstructions. Fig: This 3D reformation of the aortic arch is degraded by motion artifact from source image

57. Artifacts Stair step artifacts appear around the edges of structures in MPR and 3D reformatted images when wide collimations and non-overlapping reconstruction intervals are used. They are less severe with helical scanning, which permits reconstruction of overlapping sections. Faint stripes may be apparent in MPR and 3D reformatted images from helical data because the helical interpolation process gives rise to a degree of noise inhomogeneity along the z axis.

58. Artifacts Pulsation artifacts most commonly affect the ascending thoracic aorta and are well known to generate artifacts that can mimic aortic dissection on all images . Knowledge of this artifact and careful review of the axial images can usually distinguish a pulsation artifact from a true dissection.

59. Summary Reformation is an image rendering process in different planes and orientation from image data. For CT image reformation, all the source images must have an identical DFOV, image center, gantry tilt and they must be contiguous MPR creates coronal, sagittal and oblique images from the stack of contiguous axial slices. CPR is helpful on delineating whole structures which are tortuous and curved in a single image Intensity projection rendering is a visualization technique to enhance and display specific features or structure either of high intensity (MIP), lowest intensity (MinIP) and average intensity (AIP) in 3D dataset VRT images are considered as superior 3D technique providing more comprehensive understanding of the structures and their spatial relationship.

60. Reference Computed tomography Physical Principles, Clinical Application and Quality Control by Euclid Seeram 4 th edition Computed Tomography for Technologists – A Comprehensive Text by Lois E. Romans Jiang Hsieh - Computed Tomography Principles, Design, Artifacts, and Recent Advances-Wiley (2009) ScienceDirect researchgate.net siemens-healthineers.com/ epos.myesr.org

61. THANK YOU!!!